Gemma 4 31B (Reasoning)+1 tiers
Gemma 4 31B (Reasoning) is developed by Google, released 2026-04-02.
Reasoning-effort tiers merged. This page shows the
highest-intelligence tier. It merges 2 tiers of the same model:
Gemma 4 31B (Reasoning), Gemma 4 31B (Non-reasoning).
Input price
—
Output price
—
Blended price
—
Intelligence
14.7
65th percentile of 554 models with an Intelligence score
Coding
43.4
58th percentile of 194 models with a Coding score
Math
—
Speed
35 tok/s
Time to first token
0.92s
No API price is published for this model in our data source, so we show
— rather than guess. It may be unreleased via API, open-weights only, or simply not tracked yet.
See all of Google's models on its provider page, or browse the full model directory.